Modeling the Phase-Change Memory Material, Ge2Sb2Te5, with a Machine-Learned Interatomic Potential

Mocanu, Felix C.; Konstantinou, Konstantinos; Lee, Tae Hoon; Bernstein, Noam; Deringer, Volker L.; Csänyi, Gábor; Elliott, S. R.

doi:10.1021/acs.jpcb.8b06476

Cited by 130 publications

(135 citation statements)

References 67 publications

Supporting

Mentioning

129

Contrasting

Unclassified

Order By: Relevance

“…First, looking at Ge atoms, we see that the degree of bond alignment substantially increases for those 4 th neighbor bonds that have intermediate (around 3 Å) values. These bonds lengths are absent in tetrahedral Ge motifs and are observed in 4-fold octahedral Ge atoms (and observed in GeTe [30][31][32][33] and GeSbTe glasses [34][35][36] ). We see that these bonds are at the origin of the metallization upon excitation, as their effective charge become strongly anomal (they diverge when the gap closes).…”

Section: Resultsmentioning

confidence: 90%

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Raty

Noé

2020

Physica Rapid Research Ltrs

View full text Add to dashboard Cite

The Ovonic Threshold Switching (OTS) phenomenon, a unique discontinuity of conductivity upon electric-field application, has been observed in many chalcogenide glasses, some of which being presently used as selector elements in latest ultimate phase-change memory devices. In this work, Ab Initio Molecular Dynamics is used to simulate the structure of two prototypical glasses that have been shown to exhibit significantly different OTS properties and switching performance in OTS devices. The first glass, Ge30Se70 has a typical structure of connected Ge tetrahedra, whereas in the second Ge30Se70-based glass that contains antimony and nitrogen, the structure around Ge atoms is quite more complex. By the simulation of the excitation of electrons in the conduction band, slight modifications of the local order are shown to be sufficient to delocalize electronic states. The electron delocalization involving both Ge, Se (as well as Sb atoms in case of Sb-containing glass) ensures the percolation of conductive paths for electrons giving therefore to the excited material a metallic behavior. These conductive channels result from the local formation of 'metavalent' bonds [1, 2] in the amorphous structure as characterized geometrically and with associated Born effective charges.

show abstract

Section: Resultsmentioning

confidence: 90%

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Raty

Noé

2020

Physica Rapid Research Ltrs

View full text Add to dashboard Cite

show abstract

“…Typical DFT-MD simulations in this area encompass on the order of 300-400 atoms, [107,125,126] having very recently reached up to 900 atoms in one instance, [127] but this is only possible with fast supercomputers and sophisticated algorithms. [128] In contrast, an ML potential readily makes system sizes of many thousand atoms accessible, as we show for a structural model of bulk amorphous (a-) Ge 2 Sb 2 Te 5 , containing 7200 atoms (Figure 3c), [129] and of a partially melted GeTe nanowire, described by an even more complex structural model with over 16 000 atoms (Figure 3d). [124] An NN potential for GeTe was introduced by Sosso et al in 2012 already.…”

Section: Wwwadvmatde Wwwadvancedsciencenewscommentioning

confidence: 81%

“…[118] Copyright 2015, Wiley-VCH. [129] The small cell is one that is accessible to DFT-MD; the larger is one that is accessible to ML potentials; both are drawn to scale. [111] Reproduced with permission.…”

Section: Structure Dynamics and Function In Phase-change Materialsmentioning

confidence: 99%

See 1 more Smart Citation

Machine Learning Interatomic Potentials as Emerging Tools for Materials Science

2019

Self Cite

View full text Add to dashboard Cite

In pursuit of this goal, atomic-scale computer simulations have long been a central approach, and two major families of methods are routinely used today. On the one hand, there are quantum-mechanical simulations, in which we solve Schrödinger's equation for the electronic structure of molecular and periodic systems, most widely based on density-functional theory (DFT). [6][7][8] These methods provide (largely) reliable results for structural models of materials that normally contain a few tens or hundreds of atoms. State-of-the-art DFT methods can be applied to many material classes, and they are increasingly used for high-throughput screening and "in silico" (computer-based) design of materials: new compositions and previously unknown structures have been identified in DFT searches and subsequently experimentally realized. [5,[9][10][11] On the other hand, interatomic potential models ("force fields"), parameterizing interactions between atoms with (relatively) simple functional forms, are widely used in materials science to describe matter in molecular dynamics (MD) simulations. These simulations grant access to larger time and length scales, reaching system sizes of up to hundreds of thousands of atoms. [12] In parameterizing these potentials, a certain physical form of the atomic interactions is assumed, often in terms of bond distances, angles, and so on, and physical properties such as equilibrium lattice parameters or elastic constants enter the fitting of the potential. For this reason, such potentials are often called "empirical." They are several orders of magnitude faster than DFT, but necessarily less accurate and less easily transferable.In this Progress Report, we highlight recent developments in "machine-learned" interatomic potentials, which represent a rapidly growing field that promises to do away with the aforementioned trade-offs. Over the last year, there has been a surge of interest in machine learning (ML) methodology: part of it is due to the dramatic growth of ML throughout the scientific disciplines, and part of it is due to tangible success stories of ML-based interatomic potentials that are now beginning to emerge. We will argue that this is an exciting development with very practical implications, currently on the verge of moving from a somewhat specialized new technology to everyday applicability, poised to enhance and complement the communities' existing strengths in computational materials modeling. We will show selected applications of ML potentials to problems in materials science, discuss the current limitations (and possible pitfalls), and outline what we expect to be interesting directions for the development of the field in the coming years.Atomic-scale modeling and understanding of materials have made remarkable progress, but they are still fundamentally limited by the large computational cost of explicit electronic-structure methods such as density-functional theory. This Progress Report shows how machine learning (ML) is currently enabling a new degree of realism in materi...

show abstract

“…Ge2Sb2Te5 have been performed very recently by means of a ML-based interatomic potential [35] based on Gaussian approximations (the so-called GAP approach [18]): a representative result is reported in Figure 3e. Here, we have illustrated some of the results we have obtained by means of a NNP for GeTe.…”

Section: Crystal Nucleation and Growthmentioning

confidence: 99%

Harnessing machine learning potentials to understand the functional properties of phase-change materials

Sosso

Bernasconi

2019

MRS Bull.

View full text Add to dashboard Cite

show abstract

Modeling the Phase-Change Memory Material, Ge₂Sb₂Te₅, with a Machine-Learned Interatomic Potential

Cited by 130 publications

References 67 publications

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Machine Learning Interatomic Potentials as Emerging Tools for Materials Science

Harnessing machine learning potentials to understand the functional properties of phase-change materials

Contact Info

Product

Resources

About

Modeling the Phase-Change Memory Material, Ge2Sb2Te5, with a Machine-Learned Interatomic Potential

Cited by 130 publications

References 67 publications

Ovonic Threshold Switching in Se‐Rich GexSe1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Ovonic Threshold Switching in Se‐Rich GexSe1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Machine Learning Interatomic Potentials as Emerging Tools for Materials Science

Harnessing machine learning potentials to understand the functional properties of phase-change materials

Contact Info

Product

Resources

About

Modeling the Phase-Change Memory Material, Ge₂Sb₂Te₅, with a Machine-Learned Interatomic Potential

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism

Ovonic Threshold Switching in Se‐Rich Ge_xSe_1−x Glasses from an Atomistic Point of View: The Crucial Role of the Metavalent Bonding Mechanism